Adapting Cognitive Interview Techniques for use in Pretesting
Spanish Language Survey Instruments
Patricia Goerman
KEY WORDS: Cultural issues in survey design; data quality; non-English speaking populations
ABSTRACT
Pretesting methods that are currently used in survey research to develop and improve survey questions have been evolving for the past 20 years. While there has been increased focus on issues related to multicultural and multilingual survey design in recent years, (e.g. Harkness, Van de Vijver, and Mohler (eds.) 2003) there has been relatively little research on the appropriateness of specific pretesting methods with respondents from different cultural and/or linguistic groups.
Previous research has shown that some respondents have a great deal of difficulty with common cognitive interview probes and techniques. English-speaking respondents with low educational levels have been shown to have difficulty with both paraphrasing and think aloud probes (Willis, 2005; Bickert & Felcher, 1996; Wellens, 1994). Both Spanish and Chinese-speaking respondents have been shown to experience even greater difficulty with translated cognitive interview probes and techniques, including the use of think aloud, paraphrasing, process oriented probes, and meaning oriented probes (Pan, 2004; Carrasco, 2003; Potaka and Cochrane, 2002, 2004; Coronado and Earle, 2002; Blumberg & Goerman, 2000; Kissam, et al., 1993). In some cases these difficulties have been observed regardless of the educational level of respondents.
This paper presents an approach for continuing the process of tailoring and refining cognitive interview techniques to enhance their effectiveness across cultural and linguistic groups, with a focus on Spanish-speaking respondents. The paper discusses a plan of research for:
-
Evaluating common cognitive interview techniques and probes translated into Spanish.
-
Developing and testing alternative cognitive interview techniques for use with Spanish-speaking respondents.
Finally, the paper presents preliminary results and findings based on an implementation of this approach.
CITATION:
